Clear answers about AI and your scientific data
Understand how SciSure AI capabilities access and use information, where responsibilities differ, and what your organization should evaluate before activation.
Different AI capabilities require different controls
Model Context Protocol provides a standardized way for compatible AI environments to request information from SciSure Research.
Frequently asked questions
For detailed information about SciSure’s security controls, policies, and certifications, visit the SciSure Trust Center.
SciSure does not use customers’ proprietary experimental content to train SciSure AI models.
Data handling can differ when a capability uses an external AI provider or is developed by a Marketplace partner. Customers should review the product-specific information and applicable terms before enabling a capability.
No. SciSure AI capabilities have different purposes and access boundaries. Enabling one capability does not automatically enable other AI products or give them access to the same information.
This depends on the capability.
Astra Iris uses SciSure product documentation and does not access customer research content. SciSure MCP follows the permissions of the authenticated user and provides read-only access to supported information. Other SciSure or partner-developed add-ons may have different requirements.
No single answer applies to every capability. Some capabilities do not access customer research data, while others require information to be submitted or retrieved to perform their function.
SciSure should clearly communicate the relevant data flow before a customer activates a capability.
Processing locations depend on the selected capability, provider, deployment method, and customer environment.
Organizations with specific residency or infrastructure requirements should confirm the data flow for the individual capability before activation.
No. Astra Iris generates answers from official SciSure product documentation. It does not access customer experiments, records, or research content.
The AI Protocol Generator uses the information a user enters into the prompt to create an editable protocol draft.
Users should not include sensitive or confidential information because the AI engine operates outside the SciSure environment. All generated protocols must be reviewed and validated before use.
SciSure MCP can provide read-only access to supported SciSure Research information, including profiles, experiments, samples, sample types, and protocols.
The available information remains subject to the permissions of the authenticated user.
No. The current SciSure MCP capability is read-only. Connected AI environments cannot use it to edit experiments, samples, protocols, or other records.
Yes. When a user asks a question, relevant permitted information is returned to the connected AI environment so it can present or summarize the response.
The organization should evaluate the selected AI provider and connection method before enabling access.
Eligible customers can connect a compatible locally hosted AI model directly to their SciSure environment. When configured appropriately, this can allow prompts, retrieved information, and responses to remain within infrastructure controlled by the organization.
A direct local setup depends on the customer’s model, network, and technical configuration.
No. SciSure AI capabilities are optional and require the relevant add-on, connection, or configuration to be enabled.
Activation requirements differ by capability.
For SciSure MCP, yes. Requests follow the access permissions of the authenticated user. Information a user cannot access in SciSure Research cannot be retrieved through MCP.
Other capabilities may use different access models, which should be documented on their individual pages.
Optional add-ons and connections can be disabled or removed according to their configuration and management process. Administrators should confirm the specific procedure for the capability they are evaluating.
No. SciSure MCP is currently available for private-cloud and on-premises environments. It is not available for shared-cloud environments.
Not always. The SciSure Marketplace includes both SciSure-developed capabilities and solutions developed by partners.
Each listing identifies the developer responsible for the solution.
Not necessarily. Partner-developed add-ons may use their own infrastructure, models, subprocessors, and data-processing methods.
Customers should review the individual Marketplace listing, documentation, contractual terms, and security information before enabling a partner-developed capability.
No. Availability through the Marketplace does not remove the need for the customer to evaluate the add-on against its scientific, security, privacy, regulatory, and procurement requirements.
No. AI-generated information may be incomplete, inaccurate, or unsuitable for a particular scientific context.
Users should review outputs against trusted records, validated methods, and relevant organizational requirements.
No. AI can assist with retrieving information, drafting content, and supporting defined workflows, but qualified people remain responsible for scientific interpretation, validation, and decisions.
No. The AI Protocol Generator creates an editable starting point. Researchers must review and validate the content before using it in laboratory work.
SciSure is developing its AI management practices in alignment with ISO/IEC 42001, the international standard for artificial intelligence management systems. This includes establishing processes for AI governance, risk management, transparency, accountability, and continual improvement.
ISO/IEC 42001 establishes requirements for creating, maintaining, and continually improving an AI management system.
No technology alone makes an organization compliant. Compliance depends on how the capability is configured, governed, validated, and used within the organization’s wider processes.
The SciSure Trust Center provides current information about our security and privacy practices, compliance program, certifications, and approach to protecting customer information. Supporting documentation may also be available through the Trust Center.
Visit the SciSure Trust Center
Depending on the use case, evaluation may involve:
- Scientific or laboratory leadership
- IT and information security
- Data owners
- Privacy or legal teams
- Quality and compliance
- Procurement
- Intended users
The review should consider both the capability itself and the information users may provide to it.
Have a question about your AI requirements?
Talk with our team about the capability you are evaluating, the information involved, and the deployment or connection options available.